Generative neural networks are taking on a new role — personal shopping assistants. Analysts at Carambola Labs studied over 10 million queries to leading language models (ChatGPT, Gemini, GigaChat, Alice, Perplexity, and Grok) and found out how AI promotes products to residents of Moscow, Kazan, Nizhny Novgorod, and Krasnodar.
The study showed that algorithms don't just provide informational answers; they construct commercial offers with direct links to stores. The system switches to "seller mode" as soon as it detects trigger words in a query like "buy," "order," "find," "recommend," as well as any mention of a budget or city.
Neural networks even respond to conversational phrasing like "gift for an 8-year-old son who loves construction sets." In such cases, the bots clarify details and provide ready-made product cards with prices and specifications gathered from marketplaces. This turns an ordinary conversation into a personalized storefront, saving the buyer from lengthy searches and giving brands access to a "hot" audience.
User preferences and algorithm logic vary noticeably by region.
In Moscow, AI focuses on status: clothing and accessories dominate 72% of selections, gadgets and tech in 68%, and premium cosmetics in 53% of responses.
Kazan emphasizes technology and self-care. Algorithms recommend electronics in 72% of cases, and men's cosmetics and fragrances in 61%. Clothing lags behind significantly at 48%.
Nizhny Novgorod shows a balance: tech and fashion share the top spot (53% each), with cosmetics following closely at 51%.
Krasnodar demonstrates a more conservative approach: cosmetics (43%), tech (40%), and clothing (39%) form the core of recommendations.
"The emergence of product storefronts within AI chats is creating a new reality for e-commerce," comments Stanislav Shcherbakov, CEO of Carambola Labs. "Neural networks are gradually siphoning traffic from online stores and marketplaces. We found that algorithms take information directly from product listings. This opens up access to a new audience for brands."
Today, it's crucial for companies to optimize their product listings not only for the internal systems of trading platforms but also to implement GEO/AEO optimization. If artificial intelligence can clearly interpret a product's features and context, it will seamlessly integrate it into search results precisely when a user makes a relevant query.